Case · AI
An agentic sales system. The same results with a team 66% smaller.
Selling to governments depends on timing that no CRM tracks: elections, budget cycles, strategies ending, tenders. I built a system that watches those signals in 792 places, scores who to talk to and when, and drafts the work, with a person approving every message.
The situation
A perfect-fit government with an enthusiastic champion cannot buy if its budget closed last month or an election is eight weeks away. A lukewarm one becomes urgent the day a new minister takes office with a mandate and money to spend.
None of that timing lived anywhere. Sellers prospected on instinct and rediscovered the same political calendars again and again.
The approach
Treat political and budget timing as sales data, as structured as company size or job title. Then let agents do the research and the drafting, and keep a person on every decision that leaves the building.
What I built
- Signals for 792 places: leadership and election dates, budget cycles, news, public tenders from EU, national and development-bank portals, and where each strategy and budget sits on the path from plan to contract.
- Scoring: every contact ranked by seniority, mandate and fit, so each place has a short list of the right people. When someone changes job, the seat they left is flagged to refill.
- A daily queue, the Opportunity Radar, showing which places just became worth a conversation, and why.
- An agent team for enrichment, priority analysis, email and proposal drafts, meeting notes and CRM updates, each checked by a quality step. Six AI agents work in parallel, each owning one part of the system.
- Gates: everything that writes produces a draft a person approves, and a check on recent contact runs before any draft is written.
- A dashboard for leadership, regional sales and sales operations, with a personal access code for each person and the data encrypted.
Results
- 66%
- smaller sales team, with the same results
- 792
- places watched: 71 nations, 170 states and 551 cities
- 6
- AI agents working in parallel, each owning one part of the system
Related
01 · AIAI advisor for companies that need it to actually work